A Hybrid Method for Activity Monitoring Using Principal Component Analysis and Back-Propagation Neural Network

Human activity recognition is a useful topic of research as it recognizes various human activity ultimately helps in monitoring of patients in hospitals, elderly people at home etc. In this work a hybrid method using PCA and ANN is proposed for activity classification. Principal component analysis is used to find the important features from a number of features. In this work 381 features are selected from 561 features. Selected features are used as input to the ANN based classifier to recognize different human activity. The accuracy of the proposed activity classification method is 96.8%. Hence the proposed PCA and ANN based hybrid method can be used effectively for human activity recognition.